Gif by showtime on Giphy

For those who don’t know me, I come from the design world. I started in graphic design, then moved to digital/UX design and then on to strategy and innovation.

Since the early 2010’s the design process has looked a lot like this.. do a bunch of research/analysis, create a bunch of static docs/artifacts to convey what you’ve learned, then prototype/wireframe screens to demonstrate how a digital product/experience should act, interact, look, and feel and how users could move between different elements.

In many cases the design prototypes were viewed as the source of truth and there was soooo much work put into conveying ideas/concepts/solutions to engineering teams so they could build it to spec. I can’t begin to calculate how much time I’ve spent personally doing this 😅

As AI has accelerated and the tools have improved, the following has become absolutely crystal clear:

  • The design process (specifically the conventional UX design process) is dying and MUST change. Teams should start with testable hypotheses and quick experiments and prototypes vs trying to define/perfect everything upfront.

  • Code is the source of truth, not design artifacts.

  • Designers should be working in code. They should be using Claude Code, Cursor, etc. to build rapid prototypes over drafting documents and specs. They should be creating functional deliverables instead of static ones.

The old design process is dying. But design itself? It matters more than ever.

1. The handoff is inverting

For over 10 years the design process has looked largely the same. Discover, define, design, develop. Designers made mockups in tools like Photoshop/Sketch/Figma, wrote specs, and handed them to engineers. Engineers built what was specified, designers and product people reviewed, and everyone argued because so much got lost in translation.

That entire workflow is collapsing.

On my team, I’m challenging the heck out of my strategists and designers to get comfortable pulling down code from our Github repos, making changes using Claude Code, and submitting pull requests (for the uninitiated, this essentially means submitting a change to the code for review by someone else).

The goal is to get to the point where our engineers review those pull requests against technical architecture and quality standards, then merge to the main branch/production. Basically flipping the classic designer → developer handoff.

In the recent past, design prototypes served as the source of truth and code was trying to match. Everyone needs to flip that mental model. Code is the source of truth, and design should be influencing where code goes by giving feedback on the real, actual product.

When people can see changes running in a real browser, with real data, within minutes of thinking about it, the feedback loop tightens dramatically. You don't need to flesh out specs and map out every single edge case when prototyping when you can just build the thing and show stakeholders, users, customers, etc.

2. Code is cheap, taste isn't

"Everything looks the freaking same." Someone on my team made this statement recently and they’re right.

AI coding tools produce competent, clean, perfectly usable interfaces. They follow patterns, maintain consistency, and ship polished components. But they all look.. the same. The same card layouts, the same spacing, the same typographic hierarchy.

Code used to be the hard part, now it's cheap. What isn't cheap is taste, creativity, and the kind of distinctiveness and judgment that makes someone feel something unique/differentiated/memorable/tasteful when they use your product. These are human skills, and they just became dramatically more valuable.

The roles disappearing aren't the ones focused on these skills. They're the ones focused on production. The pixel-pushing, the spec-writing, the document-making. If your value lives in artifacts that AI can now generate in seconds, you've got a problem. But if your value lives in understanding people, applying strategy, making unexpected connections, and building trust.. you've never been more needed.

3. What AI can't do.. understand nuance & what wasn’t said

There's a level of depth that you kind of just need to sit there with someone and listen to them. You need to watch their face when they describe a frustration. You need to catch the thing they didn't say.

Experience design, customer research, and synthesis still require the kind of human sixth sense that AI isn't replicating. Understanding why a someone gets emotional when describing a problem isn't something you can automate. Watching a user struggle with a workflow and probing to understand the root cause of that struggle requires presence and a human level of curiosity.

This is where we should be spending our time. Not making documents or managing tasks. We should be sitting with people, understanding problems deeply, and then using AI tools to rapidly test ideas and solutions.

4. People aren’t ready

A lot of folks are wrestling with this shift in the work. Many have built long careers on a process that's now being upended in real time. They're watching others build/ship things in weeks that would have taken months a couple years ago. They're being asked to learn use new tools previously considered for developers only. Some are excited, some are in denial, and some are in mourning..

I'm coming at this from the perspective of a small (anti-)consultancy where everyone wears multiple hats. The dynamics are different at a 500-person org, but the direction is the same, and the people I respect most in this space are saying the same thing.

The production layer of design and the prior way of working is dying. The strategic and human layer is becoming the entire value proposition. The folks who lean into this shift will become more impactful than they've ever been. The ones who hold on to the old process won't.

Erika Flowers wrote a great piece recently called “Zero stage to orbit” that captures what’s happening right now pretty well. And Jenny Wen’s recent episode on Lenny’s podcast (below) hits on it too. The design process is blowing up and everyone needs to shift.

Putting this into action

  1. Have your team build something real this week. Not a mockup. Something running in a browser. AI tools make this possible even for people who've never written code. The learning curve is hours, not months.

  2. Flip the QA relationship. Instead of designers/product/business people handing specs to developers/engineers, have them make changes to code with AI and let engineers review for architecture and code quality or use to understand the intent and build with proper coding guardrails. The translation layer disappears and both sides learn faster.

Wrapping up

The design process as we knew it is dead/dying. The double diamond, the sequential handoff, the “Agile” sprint process, the design/developer handoff.. it doesn't match how work can/should get done in an AI world.

But design, the real discipline of understanding people, making complex things simple, and building experiences that drive value, is alive and more important than ever. The craft didn't die. The container it lived in did.

Onward & upward,
Drew

P.s. If we haven't met yet, hello. I'm Drew Burdick, Founder and Managing Partner at StealthX. We work with brands to design and build great customer experiences that win. I share ideas weekly through this newsletter and over on the Building Great Experiences podcast. Have a question? Feel free to contact us. I'd love to hear from you.

Keep Reading